Biased Perceptions of Physiological Arousal in Social Anxiety: Understanding the Role of Objective and Subjective Physiological Arousal in the Discrepancy Between Self and Observer Perceptions of Social Performance

Social Anxiety Disorder (SAD) is characterised by an intense fear of negative social evaluation (American Psychiatric Association, 2013). It commonly emerges during childhood or adolescence and is associated with academic and occupational impairments compared to non-anxious individuals (Lampe et al., 2003; Vilaplana-Pérez et al., 2019), as well as impairments in social relationships, including lower levels of intimacy, emotional expression and self-disclosure (Sparrevohn & Rapee, 2009). Individuals with SAD commonly underestimate their social performance compared to the way observers do (i.e., there is a self-observer discrepancy; Abbott & Rapee, 2004; Cheng et al., 2017; Gallego et al., 2022; Nordahl et al., 2017; Norton & Hope, 2001; Rapee & Lim, 1992; Voncken & Bögels, 2008; Wild et al., 2008). Theoretical models suggest that this discrepancy results partially from heightened self-focused attention to internal thoughts, behaviours and physiological sensations, as well as an exaggerated perception about how visible these reactions are to others (Clark & Wells, 1995; Rapee & Heimberg, 1997). Cognitive models propose that individuals use these physiological cues to construct a mental representation of their external appearance to others, however the relative contribution of objective and subjective physiological symptoms to the construction of this mental imagery (and the resulting self-observer discrepancy) is unclear. Thus, it is unclear whether individuals with SAD show heightened focus on physiological symptoms and their visibility because they actually experience higher levels of arousal during social interactions, or whether they are merely overinterpreting normative arousal and inaccurately predicting their external visibility.

There is inconsistent evidence about whether individuals with SAD have objective social skill deficits during social interactions. Some studies find that observers rate poorer social skill performance among anxious compared to non-anxious individuals (Baker & Edelmann, 2002; Cheng et al., 2017; Thompson & Rapee, 2002; Voncken & Bögels, 2008), while others have found no difference (Abbott & Rapee, 2004; Segrin & Kinney, 1995; Voncken & Bögels, 2008). Inadequate performance of social behaviours does not necessarily reflect inadequate social skill knowledge or ability, and many factors can inhibit social skill performance (Angélico et al., 2013; Hopko et al., 2001). For example, individuals with social anxiety disorder often engage in safety behaviours in the attempt to reduce the likelihood of being negatively evaluated by others (i.e. avoiding eye contact, speaking softly), which paradoxically reduces the quality of social behaviours and interactions (Piccirillo et al., 2016). Thus, behavioural tasks are limited in their ability to conclude whether an individual has a genuine social skills deficit, a performance deficit, or both (Hopko et al., 2001; Stravynski et al., 2010). Overall, there is no consistent evidence suggesting that individuals with SAD have a social skill deficit.

In contrast to the inconsistent findings about social skill deficits, studies consistently show that individuals with SAD are more critical of their social performance relative to observers (Abbott & Rapee, 2004; Cheng et al., 2017; Gallego et al., 2022; Nordahl et al., 2017; Norton & Hope, 2001; Rapee & Lim, 1992; Segrin & Kinney, 1995; Voncken & Bögels, 2008; Wild et al., 2008). This self-observer discrepancy is consistent across analogue and clinical samples, across types of behavioural tasks, and is greater among individuals with SAD than non-anxious controls (Abbott & Rapee, 2004; Gallego et al., 2022; Nordahl et al., 2017; Norton & Hope, 2001). Thus, there is strong support for a negative and inaccurate self-perception of social performance among individuals with SAD. Cognitive models of SAD suggest that this negative self-perception, comparing a negatively distorted mental self-image against the perceived expectations of others, plays a key role in the maintenance cycle of the disorder (Clark & Wells, 1995; Rapee & Heimberg, 1997).

Individuals with SAD show heightened attentional focus on internal anxiety cues (Daly et al., 1989; Deiters et al., 2013; Kley et al., 2012; Lin et al., 2021; Mansell et al., 2003; Vriends et al., 2017; Yoon & Quartana, 2012) and use these internal cues to form mental representations of how they look to others, typically benchmarking against unrealistically high standards (Clark & Wells, 1995; Heimberg et al., 2010). This perceived social performance deficit, combined with the tendency to overestimate the likelihood and consequence of negative evaluation from others, increases and maintains social anxiety over time. Interventions for SAD commonly utilise video feedback, where individuals objectively watch recordings of themselves in social interactions and contrast their perception of their behaviour and the visibility of their perceived autonomic arousal with that of an observer’s perspective, which has been shown to reduce self-observer discrepancies (Chen et al., 2010; Harvey et al., 2000; Hirsch & Clark, 2007; Hofmann, 2007; Nilsson & Lundh, 2016).

Self-focused attention and utilisation of physiological arousal cues are proposed as key mechanisms of the self-observer discrepancy. Self-focused attention (attention focused on the cognitive, affective and physiological experience of anxiety during a social task) has been shown to mediate the relationship between social anxiety and self-observer discrepancy following public speaking (Rapee & Abbott, 2007). In addition, experimental studies show that manipulating subjective perceptions of physiological arousal also increases self-observer discrepancy and negative self-perceptions of social performance (Papageorgiou & Wells, 2002; Voncken et al., 2010; Wild et al., 2008). Together, these findings support the assertion that internal focus of attention, particularly on physiological arousal is important in understanding self-observer discrepancies. While it is clear that self-focused attention plays an important role in the mental representation of oneself, the relative contributions of objective arousal and subjective arousal are less clear.

Multiple studies have demonstrated higher subjective physiological arousal in social situations among individuals with SAD compared to those with low levels of SAD, despite little to no difference in objective physiological arousal (Anderson & Hope, 2009; Edelmann & Baker, 2002; Mauss et al., 2004; Schmitz et al., 2012; Siess et al., 2014). Individuals with anxiety may overinterpret normal physiological responses in social situations, leading them to create inaccurate mental representations. Gallego et al. (2022) found some evidence that higher objective physiological arousal (skin conductance) was associated with shorter speech duration (reflecting greater distress giving a speech), however did not control for perceptions of subjective arousal. Hofmann (2006) found that those with higher social anxiety reported less social pragmatism, including less social confidence, adaptability and awareness of social cues (i.e., self-monitoring), and that this was associated with higher physiological arousal (skin conductance) in anticipation of giving a speech. Several studies have suggested that different psychophysiological indices, specifically two of the most widely examined ones - skin conductance and heart rate, are separate physiological processes that have independent effects on the experience of anxiety during speeches, with skin conductance presumed to reflect sympathetic nervous system activation while heart rate is influenced by both the sympathetic and parasympathetic nervous system (Croft et al., 2004; Hofmann et al., 2006). Croft et al. (2004) found that anxiety during a speech is associated with increases in both heart rate and skin conductance, these effects were independent of each other. Other studies have shown differential impacts of anxiety on these indices, with elevations in skin conductance but not heart rate while giving a speech (Hofmann & Kim, 2006). Overall, it is unclear how much self-focused attention, subjective physiological arousal, and objective arousal each contribute to discrepancies between self-perception and observer perception.

This study aimed to examine the relative contributions of self-focused attention, subjective physiological arousal and objective physiological arousal to the magnitude of the self-observer discrepancy in individuals with high and low levels of social speech performance anxiety. We expected that both the high and low speech anxiety groups would rate their performance significantly lower than observer ratings (i.e., demonstrate a self-observer discrepancy) and that this self-observer discrepancy would be greater in the high anxiety compared to the low speech anxiety group. We also expected greater self-focused attention in the high, compared to low, speech anxiety group, and significantly higher subjective physiological arousal despite no significant group differences in objective physiological arousal (heart rate and galvanic skin response), evidencing an exaggerated perception of arousal. We expected that subjective physiological arousal and self-focused attention would predict the self-observer discrepancy, but objective arousal would not.

The present study was the first to analyse how attention, subjectively perceived arousal, and objective arousal separately contribute to the under-estimation of performance by individuals with social anxiety. Results indicated that the self-observer discrepancy, self-focused attention, and subjective arousal were significantly higher in the high-anxiety group compared to the low-anxiety group. However, objective measures of arousal (mean GSR and BPM) did not differ between groups. Subjective arousal and self-focused attention contributed significantly to the self-observer discrepancy, but objective physiological arousal did not.

As expected, both the high and low anxiety groups underestimated their performance compared to observers, but this self-observer discrepancy was larger in the high anxiety group compared to the low anxiety group, demonstrating a greater underestimation of performance. These results replicate a negative performance bias that is documented across both speech anxious and SAD samples and several behavioural tasks, including conversations, speeches, and role plays (Cheng et al., 2017; Gallego et al., 2022; Nordahl et al., 2017; Norton & Hope, 2001; Voncken & Bögels, 2008; Wild et al., 2008). This finding is consistent with cognitive models of SAD, which propose that negative self-perceptions of performance are a result of comparing negatively distorted mental representations of the self with unrealistic social expectations (Clark & Wells, 1995; Rapee & Heimberg, 1997). Consistent with these theoretical models and empirical finding, results also showed higher self-focused attention in the high anxiety group (Deiters et al., 2013; Mansell et al., 2003), highlighting the preferential allocation of attention to internal sensations in an attempt to monitor how they are being perceived externally by others.

Results were consistent with hypotheses, with participants in the high anxiety group reporting greater subjective perceptions of arousal than participants in the low anxiety group, however, groups did not differ on objective indexes of arousal (average BPM and GSR) during the speech. There was a significant difference in heart rate between the high- and low-anxiety groups at baseline, with the high-anxiety group exhibiting elevated heart rate. However, there was no corresponding difference in skin conductance. This may reflect reduced parasympathetic regulation (i.e., reduced vagal tone) in the high anxiety group given that heart rate is modulated by both sympathetic and parasympathetic nervous system influences, while skin conductance is primarily influenced by only sympathetic nervous system activity (Croft et al., 2004; Hofmann et al., 2006). This is consistent with other studies that have found altered resting-state heart rate variability in individuals with anxiety disorders (Chalmers et al., 2014). Alternatively, the higher heart rate at baseline may suggest anticipatory anxiety in high anxious participants during the baseline component of the session given they were aware that they would be required to give a speech. Thus, it is possible that the baseline period did not truly reflect a ‘resting state’. Regardless, uncontrolled and controlled analyses suggested no difference in heart rate or skin conductance arousal between groups during the speech task. Overall, these results highlight anxiety-related differences in perceptions of physiological arousal, with individuals with social anxiety perceiving higher levels of physiological arousal than low anxious individuals, despite there being no difference in the level of objective arousal. This finding of higher levels of perceived arousal within participants with social anxiety replicates previous studies that measured both perceived and objective arousal in analogue and clinical samples of SAD (Anderson & Hope, 2009; Edelmann & Baker, 2002; Mauss et al., 2004, 2004; Schmitz et al., 2012; Siess et al., 2014). Results suggested a bidirectional misperception of physiological arousal. Those with social anxiety overestimated their arousal, while non-anxious individuals underestimated it. This may have a protective effect on self-image for those with low anxiety, while exacerbating negative self-perceptions in those with social anxiety.

This study extended previous findings (Gallego et al., 2022; Mansell & Clark, 1999; Papageorgiou & Wells, 2002; Wild et al., 2008) by examining the relative contribution of self-focused attention, subjective arousal and objective arousal in predicting the self-observer discrepancy. Results suggested that self-focused attention and subjective arousal, but not objective arousal predicted self-observer discrepancy scores. This is consistent with previous literature showing that self-focused attention influences the under-estimation of performance compared to engaging in an externally focused attention (McManus et al., 2008; Rapee & Abbott, 2007), and that the exaggerated perception of physical anxiety symptoms can also contribute to a greater underestimation of performance (Mansell & Clark, 1999; Papageorgiou & Wells, 2002; Wild et al., 2008), regardless of actual objective arousal which was not elevated in those with social anxiety. These findings are also consistent with cognitive models of SAD (Clark & Wells, 1995; Rapee & Heimberg, 1997) suggesting that the increased focus on, and perception of, physiological symptoms during stressful social situations can leads to an inaccurate mental representation of oneself, but suggests that anxious individuals do not experience objectively greater physiological arousal compared to others. Results emphasise the critical role of cognitive misinterpretation in social anxiety, while also highlighting a potentially protective underestimation of physiological arousal in non-anxious individuals. While actual physiological arousal is not elevated compared to non-anxious peers, those with SAD tend to overestimate the external visibility of normative physiological symptoms. This may explain the utility of video feedback interventions, where the inaccurate mental representations can be corrected.

There are several limitations to consider when interpreting study findings. First, observers were not blind to participant group when rating performance, thus there is potential for observer ratings of performance may have been influenced by the knowledge of participants’ anxiety level. Even if observers could have been blinded to group allocation, state anxiety ratings throughout the speech may have continued to influence observer ratings. Future studies would benefit from omitting online state anxiety ratings during speech performance and blinding raters to group. Second, the internal consistency of self-reported arousal scale (SPAS) in this sample was questionable. Given the measure assesses distinct physiological sensations, this may reflect non-uniform subjective perceptions across physiological domains. The SPAS has not been previously validated, however has been previously used for speech tasks within social anxiety samples (Mauss et al., 2004). Results warrant replication using multiple measures and samples. Third, this analogue sample was recruited on the basis of elevated public speaking anxiety given the task demands, with mean scores that indicate elevated social anxiety more broadly, results should be replicated in a sample with diagnosed social anxiety. Fourth, although group classification based on the screening items was supported by more comprehensive assessment of anxiety symptoms, one participant in the high anxiety group scored below the cut-off on the PRPSA, although sensitivity analyses suggested that results remain unchanged. Fifth, the effect of subjective arousal on the underestimation of performance may have been inflated by the timing of measurements. Heart rate and skin conductance were measured continuously throughout the speech whereas subjective arousal and self-perceived speech performance were assessed at the end of the speech. Given studies of individuals with SAD demonstrate negative post-event rumination is common, the post-task ratings may have shown more concordance than real-time assessment of objective arousal (Abbott & Rapee, 2004; Penney & Abbott, 2015). Future research would benefit from using online ratings of perceived arousal to prevent this confound from occurring. Finally, the lack of group differences in skin conductance and heart rate should be interpreted with caution given that this study did not control for other behavioural factors (e.g., vocal loudness and quantity of words spoken) that may have differed between the groups, and also contribute to sympathetic nervous system activation. While movement artifacts are likely to have been addressed given the ambulatory physiological system used (Equivital) and visual inspection of data, if these vocal behaviours were higher in the low speech anxious group, it may have obscured true group differences in activation.

This study provides novel insights into the relative contributions of self-focused attention, subjective arousal and objective arousal in the self-observer discrepancy, highlighting the key roles of self-focused attention and biased interpretations of physiological arousal cues as mechanisms influencing the underestimation of social performance. Specifically, this study emphasises the critical role of self-focused attention and cognitive misinterpretation of physiological cues in social anxiety, while reinforcing that those with social anxiety do not experience objectively elevated anxiety symptoms. Findings support the use of treatment techniques focused on correcting inaccurate perceptions and interpretations of physiological anxiety symptoms (e.g., cognitive therapy techniques including video feedback), and reducing self-focused attention (e.g., attention training), as being better aligned with these underlying mechanisms compared to those that emphasise arousal reduction (e.g., relaxation).